Free B2B data enrichment, and what credits actually buy

What does Brew do instead?

Brew reads each company's own website and DNS, then lets a model reason over what it found. Three of the four layers are free with no API key and no account, because reading a public web page is an HTTP request rather than a licensed record.

The AI layer runs on your own key, and Groq and OpenRouter both offer free models, so a few hundred rows can genuinely cost nothing. You pay your provider directly at their price with no markup from us.

There are no credits, so duplicates cost nothing and a lookup that returns nothing costs nothing. There are no export limits. There is no contract, so there is nothing to auto renew and no data to hand back. Your file stays in your browser's local storage, which also means we hold no database of your prospects, because we did not build one.

Brew is here, free and without a signup.

What Brew does not do, stated plainly

Brew is not a contact database and it will not replace one. If you arrived here searching for an alternative to a contact data provider, read this section before anything else, because five things those tools do are things we simply do not have.

We would rather lose you here than have you sign up expecting phone numbers. The reviews quoted above are not an argument that contact databases are unnecessary. They are an argument that a large part of what you currently pay per record for is company information that is sitting on the company's own website, free, and more current than any database because they maintain it themselves.

So how do the two fit together?

Use the free layers to qualify, then pay for contacts only on the rows that survive. Most teams run it the other way around, buying contact records for an entire list before anyone has checked which companies are worth approaching, and the credit complaints quoted above are partly a consequence of that order.

The sequence that wastes least.

Clean first. Remove rows with no company name. On a real 598 row list we worked with, that removed 114 rows before a single lookup ran.

Derive what you already hold. If you have an email address, the domain is inside it. A string operation gives you the website at no cost and no waiting.

Retrieve and qualify. Read the site and the DNS. Check the mail provider, because a domain with no MX record will bounce and sending to it damages the one asset in cold email you cannot buy back. Score fit on what the free layers found.

Then buy contacts. For the rows that survived, and only those. If qualification removes half your list, you have halved your credit consumption without changing provider.

This is also the honest answer to “is Brew an alternative to a contact database”. It is not. It is the step that should happen before one, and most teams skip it.

What does this actually cost?

Nothing for the website, DNS and deep scan layers. For the AI layer you bring your own key, and with a free model from Groq or OpenRouter a few hundred rows cost nothing at all. We do not sell credits, seats or a plan.

To be concrete about the comparison, using figures reviewers themselves published. One Capterra reviewer describes moving from 15,000 dollars a year with ZoomInfo to under 100 dollars a month. Another notes a ZoomInfo minimum contract around 5,000 dollars. Those are their numbers, not ours, and your quote will differ.

What we would genuinely say: if your spend is in that range and most of what you extract is company level information, the free layers will cover a meaningful share of it. If you depend on direct dials and verified personal emails, they will not, and no amount of website reading changes that.

Brew is free because it is how people find us. We are an engineering and go-to-market firm and the tool is the most honest advertisement we could build. That is the whole business model and we would rather state it than imply some other motive.

What are the real limits?

Sites with broken certificates or bot protection come back unreachable, which on a test list of 484 companies was roughly two percent. AI estimates of size and revenue are a model reading a website, not a filing. And there is no person level data at all.

Two of those are worth expanding, because they are the ones that will bite you.

Unreachable sites. We report them as unreachable rather than inventing a result. Worth noting that a company whose own site has an expired certificate is itself a signal, depending what you sell.

Inferred attributes. Employee size and revenue band are generated by a model reading page text. Treat them as sorting signals, not as facts. The difference between this and a database is that you can see the text the model read, which is more than most enrichment products show you.

We would rather publish the failure rate than have you discover it on row 300.

Why does contact data go stale so quickly?

Because it is a snapshot of something that moves. People change jobs, companies rename, offices close, and a record written eighteen months ago describes a world that no longer exists. No provider escapes this, which is why reviewers across every tool describe verifying records before using them.

The mechanics are worth understanding, because they explain which fields to trust and which to check.

Person level fields decay fastest. Job title, direct dial, personal email. These are attached to an individual who can change employer at any time, and nothing in the database learns about it until someone re-crawls or a user reports it.

Company level fields decay slowly. What a business sells, roughly how big it is, which technologies its website runs. These change over quarters rather than weeks.

Live sources do not decay at all. A company's own website is maintained by the company. When they change their pricing page, the page changes. There is no lag because there is no copy.

That is the entire argument for reading live sources where you can. It is not that databases are careless. It is that a copy of a moving thing is wrong a moment after it is made, and company websites are not copies.

The practical rule: trust a database for the things only a database has, verify anything person level before you use it in a way that matters, and get the company level fields from the company.

How do you verify an enrichment result?

Look at what the tool looked at. If a result comes with the source text, you can check it in seconds. If it comes as a bare value from a database, verification means going to the website yourself, which is the work you were paying to avoid.

Four checks that catch most errors, in the order they take least time.

Does the mail provider lookup resolve? A domain with no MX record cannot receive mail. This is binary, instant and free, and it protects your sending reputation, which is the one asset in cold outreach you cannot repurchase.

Does the site title match the stated industry? A thirty second eyeball across twenty rows catches systematic misclassification, which is the failure mode that quietly ruins a campaign.

Is the employee estimate plausible against the careers page? A company with four open roles and a team page of nine people is not the two hundred person business a model guessed.

Spot check five rows properly. Open the actual websites. Five is enough to tell whether the run is broadly right, and nobody does it, which is why bad lists reach outreach.

The broader point: enrichment you cannot audit is enrichment you have to trust. Being able to see the page text behind an inferred value is not a nice extra, it is what makes the value usable in a decision.

What should you look for in any enrichment tool?

Six things, and only one of them is coverage. Ask what happens when a lookup returns nothing, whether duplicates are charged, what the export limits are, whether you can see the source, what the contract says about termination, and whether you can leave.

What happens on an empty result. If you are charged for a lookup that returned nothing, a meaningful share of your spend buys nothing at all. Ask for the rate at which lookups come back empty on lists like yours, and treat a refusal to answer as an answer.

Duplicate handling. One reviewer quoted above describes being charged for duplicate data. Ask directly.

Export limits. A plan that caps exports at twenty five rows at a time is a plan that makes bulk work painful by design. Check before signing rather than after.

Whether you can see the evidence. Can you tell why the tool says a company is in a given industry? If not, every value is take it or leave it.

Termination terms. What happens to records you exported when the contract ends, and when does renewal trigger. A G2 reviewer raised exactly this, and auto renewal has its own community thread.

How you leave. Month to month beats annual for anything you have not yet proven. The discount for committing annually is real, and so is the cost of being locked into a tool that did not work.

None of those questions is about data coverage, which is the only thing most comparison pages measure. They are the questions that determine what the tool actually costs you.

Where does Brew sit against the newer enrichment platforms?

The newer waterfall platforms are genuinely powerful and they are a different category. They orchestrate many paid providers and charge credits for the orchestration. Brew does one narrower thing, reads live public sources, and does not charge for it.

If your workflow needs ten providers queried in sequence until one returns a mobile number, that is what those platforms are for and Brew does not compete with them. We would point you at one rather than pretend otherwise.

Where Brew fits is the step before. Most lists arriving at an enrichment platform have not been cleaned, qualified or scored, so the waterfall runs across every row including the ones that were never going to matter. Doing the free company level work first means the expensive orchestration runs on a shorter, better list.

Three practical differences worth knowing.

Scope. They cover people and companies across many sources. We cover companies from their own live sources only.

Cost model. They meter. We do not, for the free layers, and the AI layer bills you directly through your own provider key.

Where your data sits. Your list stays in your browser with Brew. That is a deliberate constraint rather than a feature, and it is also why there is no account.

We are not claiming to be a better version of a category leader. We built the part we needed, found it covered more than expected, and made it free.

What does a realistic enrichment run look like on a real list?

Messier than any demo. On a 598 row list we worked through, 114 rows were blank or date separators, 483 had an email but none had a website, and roughly two percent of sites came back unreachable. The useful work was mostly cleaning and deriving before any enrichment ran at all.

Here is what that run actually produced, because published examples in this space are almost always tidy and yours will not be.

598 rows in, 484 usable. The remainder were blank rows and date separators from however the file had been maintained. Removing them took one click and saved enriching 114 rows of nothing.

483 websites derived from email addresses, free and instant. Nobody had a website column. The domain was sitting inside the email the whole time, and a string operation produced it without a single lookup.

Real signals on the companies that resolved. One returned Webflow plus Google Analytics with Google Workspace for mail. An IT services firm returned as GitHub services and app modernisation on Webflow and Microsoft 365. A marketing agency returned WordPress and WooCommerce. A professional services firm showed an online booking widget, which tells you immediately how they take business.

About two percent unreachable, almost all expired certificates or bot protection, reported as unreachable rather than guessed at.

The honest summary of that exercise: most of the value was in the cleaning and the deriving, which cost nothing, and the enrichment ran on a list half the size it started as. That sequence is available to you whatever tools you use, and skipping it is what makes credit consumption feel unreasonable.

Does any of this apply if you already pay for a database?

Yes, and it is probably where it pays back most. Running the free company level work first shortens the list your paid tool runs against, which reduces consumption without changing provider, renegotiating anything, or giving up coverage you rely on.

Three ways teams with an existing subscription use this.

Qualify before enriching. Score fit from the company's own website, then pull contacts only for the rows that pass. If half your list drops out, you have halved consumption on the same plan.

Check deliverability before sending, not after. An MX lookup is free and instant. Sending to domains that cannot receive mail damages your sending reputation, and reputation is the one thing in cold outreach you cannot buy back at any price.

Fill the gaps your database does not cover. Technology stack, whether they publish pricing, whether they are hiring, whether they take bookings online. Several reviewers noted weaker coverage outside the United States specifically, and a company's own website does not have that problem because they maintain it themselves.

None of this requires leaving your provider, and we would not suggest you should. The argument on this page is narrower than that: a large share of what gets metered is company information that is public, current and free, and the sequence most teams run puts the expensive step first.

What is the sending reputation risk nobody prices in?

Mailing domains that cannot receive mail. Every message to a dead domain is a hard bounce, and a high bounce rate tells mailbox providers your list is unverified. Reputation damage is slow to build and slow to repair, and no budget buys it back quickly.

This is the strongest practical argument for checking the free layers before any send, and it has nothing to do with which enrichment tool you use.

A domain with no MX record cannot receive mail at all. Not slowly, not unreliably. The check takes milliseconds and costs nothing, and it removes a category of guaranteed bounce before you send.

Bounces are judged as a rate, not a count. A small list with a high bounce rate looks worse to a mailbox provider than a large list with a low one, which is why early sends on an unverified list are disproportionately damaging.

Recovery is slow. Rebuilding sender reputation takes weeks of careful sending at reduced volume. That is time your pipeline does not get back, and it is the real cost of skipping a free check.

Reviewers of paid tools describe verifying records before reaching out as routine, which tells you experienced users already treat purchased data as unverified. The question is only whether that verification happens before the send or after the bounce report.

One related habit worth adopting: knowing the mail provider also tells you what the recipient's filtering looks like, and whether a domain is on a provider that will accept your authentication setup at all.

How do you try it?

Open it and paste a list. No account, no card, no call. The free columns work immediately, and the AI columns ask for a key only when you add one.

A sensible first run on your own data takes about five minutes.

1. 01

   Upload your CSV

   A minute

   Or paste rows. It stays in your browser, so nothing is uploaded to us.

2. 02

   Clean and derive

   A minute

   Remove rows with no company name, then add a website column from the email domain if you have emails but no sites.

3. 03

   Add the free columns

   Runs while you watch

   Website signals, deep scan and mail provider. No key needed. This is the part that would consume credits elsewhere.

4. 04

   Test three rows, then run the rest

   Minutes

   Add the attribute pack if you want inference, test on three rows before running hundreds, and pick a free model if you want the whole run to cost nothing.

Open Brew. If it breaks on your list, tell us which row broke it, because that is genuinely useful to us.

• Open BrewFree, no signup, no credits
• B2B SaaS marketingPipeline, not impressions
• B2B marketing auditThe full framework, free
• Free security scanWhat your site exposes publicly
• About VoliomWho builds these